Use cases — openai prompts object to responses converter

Two concrete scenarios for the 2026-11-30 migration window. Local-only helper.

LLM platform engineer: Clear all services of hosted prompt object IDs before Nov 30

Steps: Export bodies → convert to Responses pack → PR modules → ripgrep prompt_id → retest

Result: No live callers depend on v1/prompts; ticket has conversion Markdown

Role context: llm-platform-eng in openai-prompts-nov30-shutdown. Deadline 2026-11-30. Vehicle converter-card. Keep the export attached to the ticket and re-verify official sources.

AI application owner: Move one production prompt with variables into Responses instructions/input

Steps: Paste body → map variables → generate payload sketch → code review → deploy

Result: Production call uses local instructions/input; hosted prompt retired

Role context: llm-platform-eng in openai-prompts-nov30-shutdown. Deadline 2026-11-30. Vehicle converter-card. Keep the export attached to the ticket and re-verify official sources.

Next: home tool · how-to · FAQ.

Before November 30, treat every hosted prompt id as a production dependency: owners, environments, and rollback notes belong in the same ticket as the Responses conversion pack. Keep a short list of services that still import prompt fixtures in CI so the cutover does not surprise staging. If a prompt embeds tool schemas, move those schemas beside the instructions module rather than leaving them only in the dashboard. Re-read the official deprecations table the morning of cutover in case OpenAI publishes a last-minute clarification. This page never uploads prompt text and never stores API keys.